PROBLEM STATEMENT There are tonnes of documents out there containing a lot of information. There is a need for a web-based search engine or a bot that can quickly give us the answers by reading the document.
OUR SOLUTION We have trained our model to answer questions related to a document. It can be any type of document, for ex:- documentation of a framework or a library, digital notes or study material, textbooks, novels, etc. Whenever a query is given to the model, it will provide a concise answer to the question. If asked to summarise a chapter or describe an entity, it does that too.
TECHNOLOGIES USED
- Haystack
- Google Colab
- Elastic search
- Python
FUTURE SCOPE The model has a huge scope of expansion. In terms of technology, it could be done as follows:
- Integrate this with a chatbot or a web based application in various domains.
- Integrate other services like those from MS Azure so that the model will have enhanced features.
- Use deep learning algorithms to improve the quality of output.
In terms of real life applications, following are some of the domains this could be used in:
- For working professionals, browsing through emails is a big task. This could be used to see what information has been communicated to whom.
- For developers, it can help them search through huge documentations of services.
- Help people prepare for certification courses ; one way of doing that is training the model to generate random question and answers.
- Help students browse through digital study materials.